I won but lost rating?

I meant that the mate was theoretically impossible, but practically possible, shall the other side blunder hugely.

Historical draws by clock flag has been reversed only if a forced checkmate is possible and the player is able to prove that to the arbiter. Lichess doesn't have an arbiter. This means that you'll just have to accept the draw as is.

@Fenris1066: regarding this topic, FICS is not different from lichess or any other internet chess server that I know of; they would all evaluate a draw here. This IS a deviation from the official FIDE rules, but it is probably a justifiable deviation. After all, it would be difficult to write an algorithm that can quickly determine for any given position if there exists a legal continuation that will result in a checkmate.

The other side of the coin sometimes comes up too: an ending that should be judged as a draw, but is instead judges as a win. This will happen for example when there is nothing on the board but pawn chains that block entry for both kings, meaning checkmate is impossible. I have seen this happen in an OTB blitz game where one player ran out of time and the arbiter mistakenly awarded a win to the other player.

Unihedron: This is the exact opposite of what the words "theoretically" and "practically" mean. :-)

@erindreki what I meant about FICS is that if you have a position that is judged a draw by the algorithm, FICS has an arbitration system and you could appeal it and get the result changed.

The reason that isn't done here is that for lichess that's a big hassle because there isn't an algorithm in place for adjusting ratings after the fact that then takes into account all the games played after, all their opponents, all their opponents opponents and so on. Changing one person's rating by whatever slight difference that game would be unfair because it should in theory change all the ratings of all the games related that followed it. If the arbitration takes 2 or 3 days and there is a really active player in that chain, it would get absurd. So the compromise was reached to just do like most other online sites and implement the insufficient material check, with what flaws it may have, to make it more fair. The cases in which it's wrong are a very small percent of the games that a timeout occurs when there is not a practical way to win.

Also, lichess doesn't use the FIDE rules, it's not a FIDE sponsored site. So the FIDE rules (along with other chess organizations' rules) are simply guidelines here.

And #11, erindreki is correct. You're using the words backwards.

Theoretically possible means you can show a situation in which it could occur, even if it is unlikely (i.e. requires the opponent to play inaccurately).

Practically possible means the position has a good chance of being reached under precise play.

#10: Your information about FICS here is incorrect, possibly you are confusing this with their system for adjudicating stored games. Per discussion with an FICS admin:

"Rochester(*)(SR)(TM)(1): No, a lone minor piece is always considered insufficient mating material here."